revert awq dtype change for vllm inference limitation#1613
Merged
WeiweiZhang1 merged 5 commits intomainfrom Mar 26, 2026
Merged
revert awq dtype change for vllm inference limitation#1613WeiweiZhang1 merged 5 commits intomainfrom
WeiweiZhang1 merged 5 commits intomainfrom
Conversation
Signed-off-by: WeiweiZhang1 <weiwei1.zhang@intel.com>
Contributor
There was a problem hiding this comment.
Pull request overview
This PR aims to improve robustness of vLLM AWQ inference across different CUDA devices by ensuring the exported model’s dtype metadata aligns with vLLM’s AWQ kernel limitations.
Changes:
- Force
torch.float16dtype metadata during AWQ export to improve vLLM compatibility. - Extend AutoRound export dtype selection to prefer FP16 when the packing format is AWQ.
- Update the vLLM AWQ integration test to pass an explicit
dtypeargument toLLM(...).
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
test/test_cuda/integrations/test_vllm.py |
Adjusts vLLM initialization for the AWQ integration test. |
auto_round/export/export_to_awq/export.py |
Forces AWQ exports to write FP16 dtype metadata via save_model(..., dtype=...). |
auto_round/export/export_to_autoround/export.py |
Selects FP16 dtype metadata when the packing format indicates AWQ. |
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
xin3he
approved these changes
Mar 25, 2026
Contributor
|
/azp run Unit-Test-CUDA-AutoRound |
|
Azure Pipelines successfully started running 1 pipeline(s). |
wenhuach21
reviewed
Mar 25, 2026
Signed-off-by: WeiweiZhang1 <weiwei1.zhang@intel.com>
Contributor
|
/azp run Unit-Test-CUDA-AutoRound |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Signed-off-by: WeiweiZhang1 <weiwei1.zhang@intel.com>
Contributor
|
/azp run Unit-Test-CUDA-AutoRound |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Contributor
|
/azp run Unit-Test-CUDA-AutoRound |
|
Azure Pipelines successfully started running 1 pipeline(s). |
WeiweiZhang1
deleted the
revert_awq_dtype_change_for_vllm_infer_limitation
branch
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
The performance of vllm awq inference varies across different devices; the CUDA restrictions on the float16 data type still apply on the A100, so revert the data type change to ensure the robustness of the inference.
Type of Change
Related Issues
Fixes or relates to #
Checklist Before Submitting